federated_learning_Maths_Handwritten.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Saugatkafley/Federated-Learning/blob/main/federated_learning_Maths_Handwritten.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tL1cV9R2Ymmo"
      },
      "source": [
        "## Mounting Kaggle Dataset Maths Handwritten.\n",
        "At first you have to upload the `kaggle.json` file , which can be obtained from `kaggle.com` -> `Account Settings`.\n",
        "Scroll a bit and there obtain `Create New API token`. \n",
        "Then upload it to the directory."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "89TIcP1CYfLX"
      },
      "outputs": [],
      "source": [
        "!pip install kaggle"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rU8sVujueF-y"
      },
      "source": [
        "Making a directory --> `./kaggle` \n",
        "\n",
        "Copying `kaggle.json` to the directory\n",
        "\n",
        "Changing the permisssion to `600`."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Vw8EJD5GZltR"
      },
      "outputs": [],
      "source": [
        "! mkdir ~/.kaggle\n",
        "! cp kaggle.json ~/.kaggle/\n",
        "! chmod 600 ~/.kaggle/kaggle.json"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "QolI7FIgAWTg"
      },
      "outputs": [],
      "source": [
        "# ! kaggle datasets download -d sagyamthapa/handwritten-math-symbols\n",
        "! kaggle datasets download -d xainano/handwrittenmathsymbols\n",
        "# !unzip mnistasjpg.zip\n",
        "! unzip handwrittenmathsymbols.zip\n",
        "! sudo apt install unrar\n",
        "! unrar x data.rar /content/data/"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8xSJL0mpC30I",
        "outputId": "ca3eb5fd-e4df-41b8-e60c-112a53559a09"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting split-folders\n",
            "  Downloading split_folders-0.5.1-py3-none-any.whl (8.4 kB)\n",
            "Installing collected packages: split-folders\n",
            "Successfully installed split-folders-0.5.1\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "Copying files: 375974 files [01:12, 5167.81 files/s]\n"
          ]
        }
      ],
      "source": [
        "! pip install split-folders\n",
        "\n",
        "import splitfolders\n",
        "\n",
        "input_folder  = '/content/data/extracted_images/'\n",
        "\n",
        "splitfolders.ratio(\n",
        "    input_folder,\n",
        "    output = \"dataset\",\n",
        "    seed =42,\n",
        "    ratio = (.2,.8),\n",
        ")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ekGOiwyNO8mF"
      },
      "source": [
        "## Importing Libraries "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pXb8S1S2YajE",
        "outputId": "22c15f33-7527-491b-a7e5-d709ed16aab5"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: imutils in /usr/local/lib/python3.7/dist-packages (0.5.4)\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import random\n",
        "import cv2\n",
        "import os\n",
        "from tqdm import tqdm\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import LabelBinarizer\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.utils import shuffle\n",
        "from sklearn.metrics import accuracy_score\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow import expand_dims\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.layers import Conv2D, Input, Lambda\n",
        "from tensorflow.keras.layers import MaxPooling2D\n",
        "from tensorflow.keras.layers import Activation\n",
        "from tensorflow.keras.layers import Flatten\n",
        "from tensorflow.keras.layers import Dense\n",
        "from tensorflow.keras.optimizers import SGD\n",
        "from tensorflow.keras import backend as K\n",
        "\n",
        "!pip install imutils\n",
        "from imutils import paths"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "id": "r950km2A-l0w",
        "outputId": "6c37e1f7-66d0-47f6-a466-519ade52e565"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# import cv2\n",
        "# import matplotlib.pyplot as plt\n",
        "\n",
        "# # img  = cv2.imread(\"5.png\" )\n",
        "# img  = cv2.imread(\"/content/dataset/train/sum/exp100626.jpg\")\n",
        "# # img_copy = img.copy()\n",
        "# # img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n",
        "# plt.imshow(img, cmap = 'gray')\n",
        "# plt.title(\"Image--Sum\")\n",
        "# plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qx8f_JkQEDu6"
      },
      "source": [
        "## Training Dataset Preprocessing."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "r9aOeT09Yajf"
      },
      "outputs": [],
      "source": [
        "def load(paths, verbose=-1):\n",
        "    '''expects images for each class in seperate dir, \n",
        "    e.g all digits in 0 class in the directory named 0 '''\n",
        "    data = list()\n",
        "    labels = list()\n",
        "    # loop over the input images\n",
        "    for (i, imgpath) in enumerate(paths):\n",
        "        # load the image and extract the class labels        \n",
        "        image = cv2.imread(imgpath )\n",
        "\n",
        "        # print(image.shape)\n",
        "        label = imgpath.split(os.path.sep)[-2]\n",
        "        # scale the image to [0, 1] and add to list\n",
        "        data.append(image/255)\n",
        "        labels.append(label)\n",
        "    return data, labels\n",
        "\n",
        "def create_clients(image_list, label_list, num_clients=10, initial='clients'):\n",
        "    ''' return: a dictionary with keys clients' names and value as \n",
        "                data shards - tuple of images and label lists.\n",
        "        args: \n",
        "            image_list: a list of numpy arrays of training images\n",
        "            label_list:a list of binarized labels for each image\n",
        "            num_client: number of fedrated members (clients)\n",
        "            initials: the clients'name prefix, e.g, clients_1 \n",
        "            \n",
        "    '''\n",
        "\n",
        "    #create a list of client names\n",
        "    client_names = ['{}_{}'.format(initial, i+1) for i in range(num_clients)]\n",
        "\n",
        "    #randomize the data\n",
        "    data = list(zip(image_list, label_list))\n",
        "    random.shuffle(data)  # <- IID\n",
        "\n",
        "    #shard data and place at each client\n",
        "    size = len(data)//num_clients\n",
        "    shards = [data[i:i + size] for i in range(0, size*num_clients, size)]\n",
        "\n",
        "    #number of clients must equal number of shards\n",
        "    assert(len(shards) == len(client_names))\n",
        "\n",
        "    return {client_names[i] : shards[i] for i in range(len(client_names))} \n",
        "\n",
        "\n",
        "def batch_data(data_shard, bs=32):\n",
        "    '''Takes in a clients data shard and create a tfds object off it\n",
        "    args:\n",
        "        shard: a data, label constituting a client's data shard\n",
        "        bs:batch size\n",
        "    return:\n",
        "        tfds object'''\n",
        "    #seperate shard into data and labels lists\n",
        "    data, label = zip(*data_shard)\n",
        "    # dataset = tf.data.Dataset.from_tensor_slices(list(data), list(label))\n",
        "    dataset = tf.data.Dataset.from_tensor_slices((list(data), list(label)))\n",
        "    return dataset.shuffle(len(label)).batch(bs)\n",
        "\n",
        "\n",
        "def weight_scalling_factor(clients_trn_data, client_name):\n",
        "    client_names = list(clients_trn_data.keys())\n",
        "    #get the bs\n",
        "    bs = list(clients_trn_data[client_name])[0][0].shape[0]\n",
        "    #first calculate the total training data points across clients\n",
        "    global_count = sum([tf.data.experimental.cardinality(clients_trn_data[client_name]).numpy() for client_name in client_names])*bs\n",
        "    # get the total number of data points held by a client\n",
        "    local_count = tf.data.experimental.cardinality(clients_trn_data[client_name]).numpy()*bs\n",
        "  \n",
        "    return local_count/global_count\n",
        "\n",
        "\n",
        "def scale_model_weights(weight, scalar):\n",
        "    '''function for scaling a models weights'''\n",
        "    weight_final = []\n",
        "    steps = len(weight)\n",
        "    for i in range(steps):\n",
        "        weight_final.append(scalar * weight[i])\n",
        "    return weight_final\n",
        "\n",
        "\n",
        "\n",
        "def sum_scaled_weights(scaled_weight_list):\n",
        "    '''Return the sum of the listed scaled weights. The is equivalent to scaled avg of the weights'''\n",
        "    avg_grad = list()\n",
        "    #get the average grad accross all client gradients\n",
        "    for grad_list_tuple in zip(*scaled_weight_list):\n",
        "        layer_mean = tf.math.reduce_sum(grad_list_tuple, axis=0)\n",
        "        avg_grad.append(layer_mean)\n",
        "        \n",
        "    return avg_grad\n",
        "\n",
        "\n",
        "def test_model(X_test, Y_test,  model, comm_round):\n",
        "    cce = tf.keras.losses.CategoricalCrossentropy(from_logits=True)\n",
        "    logits = model.predict(X_test)\n",
        "    loss = cce(Y_test, logits)\n",
        "    acc = accuracy_score(tf.argmax(logits, axis=1), tf.argmax(Y_test, axis=1))\n",
        "    print('comm_round: {} | global_acc: {:.3%} | global_loss: {}'.format(comm_round, acc, loss))\n",
        "    return acc, loss\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wINO-rQoPP9d"
      },
      "source": [
        "## Building CNN model\n",
        "Model has `input-shape` of `[45,45,3]`\n",
        "\n",
        "`Conv2D` layer having `filter` = `32` , `kernel_size` = `3` and Activation function  = `'relu'`\n",
        "\n",
        "`MaxPool2D` layer having `strides` and `pool_size`  = `2`\n",
        "\n",
        "And the after 2 `Conv2D`layers the layers are `Flatten`.\n",
        "The layers are then fully connected by `Dense` layers . \n",
        "\n",
        "The output is `Dense` layer of `82` units having activation `softmax`."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "V8sqXLqRL5Vj"
      },
      "outputs": [],
      "source": [
        "def create_keras_model(shape , classes):\n",
        "  model = tf.keras.Sequential()\n",
        "  \n",
        "  model.add(tf.keras.layers.Conv2D(filters =32 , kernel_size =3  , activation='relu' , input_shape = [shape[0], shape[1] , shape[2] ]))\n",
        "  # model.add(Lambda(lambda x: expand_dims(x, axis=-1)))\n",
        "  # model.add(tf.keras.layers.Conv2D(filters =32 , kernel_size =3  , activation='relu' , input_shape  =(shape,)))\n",
        "  model.add(tf.keras.layers.MaxPool2D( strides = 2,pool_size=2))\n",
        "\n",
        "  model.add(tf.keras.layers.Conv2D(filters = 32, kernel_size = 3 , activation = 'relu' ))\n",
        "  model.add(tf.keras.layers.MaxPool2D(strides = 2, pool_size  =2))\n",
        "\n",
        "  model.add(tf.keras.layers.Flatten())\n",
        "  model.add(tf.keras.layers.Dense(units  = 128 , activation = 'relu'))\n",
        "  model.add(tf.keras.layers.Dropout(0.25))\n",
        "  model.add(tf.keras.layers.Dense(units  = 128 , activation = 'relu'))\n",
        "  model.add(tf.keras.layers.Dropout(0.25))\n",
        "  model.add(tf.keras.layers.Dense(units  =classes  ,activation  ='softmax'))\n",
        "  return model  "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RRSFOsQMQ6uc"
      },
      "source": [
        "## Getting Image from the directories and giving them respective labels. \n",
        "\n",
        "`Labels` are then binary encoding by `LabelBinarizer()`"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6jybBZygYajq",
        "outputId": "c79709de-b4b9-4cef-f6df-a2ff795443ab"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "75161\n",
            "75161\n"
          ]
        }
      ],
      "source": [
        "#declear path to your Image data folder\n",
        "\n",
        "img_path = './dataset/train/'\n",
        "image_paths = sorted(list(paths.list_images(img_path)))\n",
        "\n",
        "#apply our function\n",
        "image_list, label_list = load(image_paths)\n",
        "\n",
        "print(len(image_list))\n",
        "#binarize the labels\n",
        "lb = LabelBinarizer()\n",
        "label_list = lb.fit_transform(label_list)\n",
        "print(len(label_list))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q7USb1bwRURl"
      },
      "source": [
        "Splitting dataset into `X_train, X_test, y_train, y_test`"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "X70EUPjLYajr"
      },
      "outputs": [],
      "source": [
        "#split data into training and test set\n",
        "X_train, X_test, y_train, y_test = train_test_split(image_list, \n",
        "                                                    label_list, \n",
        "                                                    test_size=0.1, \n",
        "                                                    random_state=42)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pwvm90BaYajv"
      },
      "source": [
        "### IID"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JQh87cnCYajw",
        "outputId": "1175b446-1a9c-4042-f91b-30ed88c80464"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(67644, 7517, 67644, 7517)"
            ]
          },
          "execution_count": 6,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "len(X_train), len(X_test), len(y_train), len(y_test)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ENsBsxe1RgLg"
      },
      "source": [
        "## Creating Clients and batching them with datasets."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "z5iyVxTsYajx"
      },
      "outputs": [],
      "source": [
        "#create clients\n",
        "num_clients = 10\n",
        "clients = create_clients(X_train, y_train, num_clients=num_clients, initial='client') #10 clients"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "2A0HsxVqYajz"
      },
      "outputs": [],
      "source": [
        "# process and batch the training data for each client\n",
        "clients_batched  = dict()\n",
        "for (client_name, data) in clients.items():\n",
        "    clients_batched[client_name] = batch_data(data)\n",
        "\n",
        "    \n",
        "#process and batch the test set  \n",
        "test_batched = tf.data.Dataset.from_tensor_slices((X_test, y_test)).batch(len(y_test))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cd17yPL1Rpwo"
      },
      "source": [
        "## Global Training Parameters"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "H_rN_WivYajz"
      },
      "outputs": [],
      "source": [
        "comms_round = 1\n",
        "loss='categorical_crossentropy'\n",
        "metrics = ['accuracy']\n",
        "optimizer  = 'adam'    "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "iH4Jd-WXYaj0"
      },
      "outputs": [],
      "source": [
        "#initialize global model\n",
        "build_shape = [45,45,3] # Train image Dimensions\n",
        "classes = 82\n",
        "global_model = create_keras_model(build_shape , classes)\n",
        "global_acc_list = []\n",
        "global_loss_list = []"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 762
        },
        "id": "wJkOPO3RYaj0",
        "outputId": "fdf35cee-25a3-47ca-fd7a-18bdfa143aa0"
      },
      "outputs": [
        {
          "ename": "ValueError",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-22-bf7cf1fc8af4>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     28\u001b[0m     \u001b[0mlocal_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mset_weights\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mglobal_weights\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     29\u001b[0m     \u001b[0;31m#fit local model with client's data\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 30\u001b[0;31m     \u001b[0mlocal_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mclients_batched\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mclient\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     31\u001b[0m     \u001b[0mloss\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlocal_model\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhistory\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     32\u001b[0m     \u001b[0mlocal_acc_list\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mloss\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py\u001b[0m in \u001b[0;36merror_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m     65\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint: disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     66\u001b[0m       \u001b[0mfiltered_tb\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0m_process_traceback_frames\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m__traceback__\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 67\u001b[0;31m       \u001b[0;32mraise\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwith_traceback\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfiltered_tb\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     68\u001b[0m     \u001b[0;32mfinally\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     69\u001b[0m       \u001b[0;32mdel\u001b[0m \u001b[0mfiltered_tb\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\u001b[0m in \u001b[0;36mautograph_handler\u001b[0;34m(*args, **kwargs)\u001b[0m\n\u001b[1;32m   1145\u001b[0m           \u001b[0;32mexcept\u001b[0m \u001b[0mException\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m  \u001b[0;31m# pylint:disable=broad-except\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1146\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mhasattr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"ag_error_metadata\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1147\u001b[0;31m               \u001b[0;32mraise\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mag_error_metadata\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mto_exception\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1148\u001b[0m             \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1149\u001b[0m               \u001b[0;32mraise\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mValueError\u001b[0m: in user code:\n\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 1021, in train_function  *\n        return step_function(self, iterator)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 1010, in step_function  **\n        outputs = model.distribute_strategy.run(run_step, args=(data,))\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 1000, in run_step  **\n        outputs = model.train_step(data)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 860, in train_step\n        loss = self.compute_loss(x, y, y_pred, sample_weight)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/training.py\", line 919, in compute_loss\n        y, y_pred, sample_weight, regularization_losses=self.losses)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/compile_utils.py\", line 184, in __call__\n        self.build(y_pred)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/compile_utils.py\", line 132, in build\n        self._losses = self._conform_to_outputs(y_pred, self._losses)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/compile_utils.py\", line 56, in _conform_to_outputs\n        struct = map_to_output_names(outputs, self._output_names, struct)\n    File \"/usr/local/lib/python3.7/dist-packages/keras/engine/compile_utils.py\", line 652, in map_to_output_names\n        'Found unexpected losses or metrics that do not correspond '\n\n    ValueError: Found unexpected losses or metrics that do not correspond to any Model output: dict_keys(['loss', 'accuracy']). Valid mode output names: ['dense_11']. Received struct is: {'loss': ListWrapper([2.1534390449523926]), 'accuracy': ListWrapper([0.4926079213619232])}.\n"
          ]
        }
      ],
      "source": [
        "#commence global training loop\n",
        "for comm_round in range(comms_round):\n",
        "          \n",
        "  # get the global model's weights - will serve as the initial weights for all local models\n",
        "  global_weights = global_model.get_weights()\n",
        "  \n",
        "  #initial list to collect local model weights after scalling\n",
        "  scaled_local_weight_list = list()\n",
        "\n",
        "  #randomize client data - using keys\n",
        "  all_client_names = list(clients_batched.keys())\n",
        "          \n",
        "  client_names = random.sample(all_client_names, k=10)\n",
        "  # print(client_names, len(client_names))\n",
        "  random.shuffle(client_names)\n",
        "\n",
        "  #loop through each client and create new local model\n",
        "  for client in client_names:\n",
        "    local_model = create_keras_model(build_shape, classes)\n",
        "    local_model.compile(loss=loss, \n",
        "                  optimizer=optimizer, \n",
        "                  metrics=metrics)\n",
        "    \n",
        "    #set local model weight to the weight of the global model\n",
        "    local_model.set_weights(global_weights)\n",
        "    #fit local model with client's data\n",
        "    local_model.fit(clients_batched[client], epochs=1)\n",
        "    \n",
        "    #scale the model weights and add to list\n",
        "    scaling_factor = 0.1 # weight_scalling_factor(clients_batched, client)\n",
        "    \n",
        "    scaled_weights = scale_model_weights(local_model.get_weights(), scaling_factor)\n",
        "    scaled_local_weight_list.append(scaled_weights)\n",
        "    \n",
        "    #clear session to free memory after each communication round\n",
        "    K.clear_session()\n",
        "    \n",
        "  #to get the average over all the local model, we simply take the sum of the scaled weights\n",
        "  average_weights = sum_scaled_weights(scaled_local_weight_list)\n",
        "  \n",
        "  #update global model \n",
        "  global_model.set_weights(average_weights)\n",
        "\n",
        "  #test global model and print out metrics after each communications round\n",
        "  for(X_test, Y_test) in test_batched:\n",
        "      global_acc, global_loss = test_model(X_test, Y_test, global_model, comm_round)\n",
        "      global_acc_list.append(global_acc)\n",
        "      global_loss_list.append(global_loss)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ratgY_rxSNZV"
      },
      "source": [
        "## Making a single prediction with Global Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4GLiBxkec1xK",
        "outputId": "38e0f032-5fa9-45b2-8a82-f638c21130c9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "4\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "from keras.preprocessing import image\n",
        "test_image = image.load_img('5.png' , target_size = (45,45))\n",
        "#change to PIL format\n",
        "test_image = image.img_to_array(test_image) \n",
        "test_image = np.expand_dims(test_image , axis = 0) #extra dimension to batch\n",
        "result = global_model.predict(test_image)\n",
        "result = lb.inverse_transform(result)\n",
        "result1 = \"\".join(str(e) for e in result)\n",
        "print(result1)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6aX-BRvVSZAt"
      },
      "source": [
        "## Saving and Loading Models"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8wTMK1pHCOOX",
        "outputId": "bb3287ac-4d4a-498e-e56f-d60d8e36e018"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "F_A3PYRLqlvX",
        "outputId": "5c471ae2-4287-419a-e3d0-2b875da3fed6"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Compiled the loaded model, but the compiled metrics have yet to be built. `model.compile_metrics` will be empty until you train or evaluate the model.\n",
            "INFO:tensorflow:Assets written to: /content/drive/MyDrive/global_model/assets\n"
          ]
        }
      ],
      "source": [
        "global_model.save(\"/content/drive/MyDrive/global_model/\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bSya1SOqCN0k",
        "outputId": "7c951ec2-6acb-412c-933e-ec35a8125378"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:No training configuration found in save file, so the model was *not* compiled. Compile it manually.\n"
          ]
        }
      ],
      "source": [
        "import tensorflow as tf\n",
        "global_model = tf.keras.models.load_model(\"/content/drive/MyDrive/global_model/\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dSbM9mXBl3gc"
      },
      "source": [
        "## Using OPENCV to classify multiple digits prediction using bounding box.\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "er9S8sJL0mzs",
        "outputId": "8cbea865-5291-418d-cef6-c5bd70431a53"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Collecting wget\n",
            "  Downloading wget-3.2.zip (10 kB)\n",
            "Building wheels for collected packages: wget\n",
            "  Building wheel for wget (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for wget: filename=wget-3.2-py3-none-any.whl size=9675 sha256=fee381913925f6d1964e2854bc31f2d2f69df3e92e03d45266f0b2a58194dccd\n",
            "  Stored in directory: /root/.cache/pip/wheels/a1/b6/7c/0e63e34eb06634181c63adacca38b79ff8f35c37e3c13e3c02\n",
            "Successfully built wget\n",
            "Installing collected packages: wget\n",
            "Successfully installed wget-3.2\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/google/colab/_pip.py:87: ResourceWarning: unclosed file <_io.TextIOWrapper name='/usr/local/lib/python3.7/dist-packages/wget-3.2.dist-info/top_level.txt' mode='r' encoding='UTF-8'>\n",
            "  for line in open(toplevel):\n",
            "ResourceWarning: Enable tracemalloc to get the object allocation traceback\n"
          ]
        }
      ],
      "source": [
        "! pip install wget"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "LcB91p4wyu7M"
      },
      "outputs": [],
      "source": [
        "import wget\n",
        "file_name = wget.download (\"https://i.imgur.com/hRmXkdQ.png\")\n",
        "# ! wget https://i.imgur.com/PwWiTk8.png\n",
        "# file_url  = \"https://i.imgur.com/PwWiTk8.png\"\n",
        "# file_name = file_url.split(\"/\")[-1]\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 240
        },
        "id": "vXWm_b8yl7Qf",
        "outputId": "59e02a7c-7416-40ed-9b4d-0f1133387c28"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "from tensorflow.keras.preprocessing import image\n",
        "import cv2\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# img  = cv2.imread(\"5.png\" )\n",
        "img  = cv2.imread(file_name)\n",
        "img_copy = img.copy()\n",
        "img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n",
        "plt.imshow(img, cmap = 'gray')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 252
        },
        "id": "uu_xktICwpEI",
        "outputId": "77902fe5-bd9c-4d8c-f47d-7418a54336db"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# img = image.img_to_array(img, dtype='uint8')\n",
        "(thresh, img_bin) = cv2.threshold(img_gray, 20, 255 , cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)\n",
        "plt.imshow(img_bin,cmap='gray')\n",
        "plt.title('Threshold: {}'.format(thresh))\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 258
        },
        "id": "7hlugLzv05Ym",
        "outputId": "78c5babc-cc45-465c-bbee-01794a5606f0"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f87d5a9a690>"
            ]
          },
          "execution_count": 132,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "cv2.floodFill(img_bin, None, (0, 0), 0)\n",
        "plt.imshow(img_bin,)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ALvABQ9c1LrH"
      },
      "outputs": [],
      "source": [
        "# Get each bounding box\n",
        "# Find the big contours/blobs on the filtered image:\n",
        "\n",
        "# Bug use img_gray not binary image \n",
        "contours, hierarchy = cv2.findContours(img_bin, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HDebl5T9pXjb"
      },
      "outputs": [],
      "source": [
        "sorted_ctrs = sorted(contours, key=lambda ctr: cv2.boundingRect(ctr)[0])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "ZXitGlj11Yod",
        "outputId": "2b27e4d4-2d7a-4825-d83d-810632ef80de"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "['3']\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<PIL.Image.Image image mode=RGB size=78x165 at 0x7F87D5D151D0>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "['3', '+']\n"
          ]
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFEAAABmCAIAAAA8kup0AAABF0lEQVR4nO3a3YrCQBAFYVvy/q8c70IEwb/u5GjVd7cgizU9TpJ1a13XC8z17DdwAps7VFVVtf/aRs3NW21ytnubwWYGmxlsZrCZwWYGmxlsbhX7CO2cv/YTf0V1zgw2M9jMYDODzd0yb7mdc4f8W27nzGAzg80DAi/RzrlJ+CXaOTMQm5f9D4Fn7ITazpsDgkPOtuX5S/rMLetbq1n7V//x3r7LnNhvmWu3lU7dk4R8dB866PM8twQv7qn9Gzj0DJvwwWoS70lsHhN1kjvnVrGXK+fMYDODzQw2M9jMcFxzzmOGc+6W+ZjhnBlsZrCZwWYGmxlsZrCZwWYGmxlsZrCZwWYGmxnGm7evb3K+xxn5n/Vw7m2GGwLAP771RUTEAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<PIL.Image.Image image mode=RGB size=81x102 at 0x7F87C4826CD0>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "['3', '+', '5']\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<PIL.Image.Image image mode=RGB size=130x137 at 0x7F87D5945950>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "8\n"
          ]
        }
      ],
      "source": [
        "from google.colab.patches import cv2_imshow\n",
        "from keras.preprocessing import image\n",
        "import numpy as np\n",
        "predictions  = []\n",
        "# Look for the outer bounding boxes (no children):\n",
        "current_crop = [] # hold the cropped images\n",
        "for _, c in enumerate(sorted_ctrs):\n",
        "\n",
        "  # Get the bounding rectangle of the current contour:\n",
        "  boundRect = cv2.boundingRect(c)\n",
        "\n",
        "  # Get the bounding rectangle data:\n",
        "  rectX = boundRect[0]\n",
        "  rectY = boundRect[1]\n",
        "  rectWidth = boundRect[2]\n",
        "  rectHeight = boundRect[3]\n",
        "  \n",
        "  # Estimate the bounding rect area:\n",
        "  rectArea = rectWidth * rectHeight\n",
        "  # print(rectArea)\n",
        "  # Set a min area threshold\n",
        "  minArea =40\n",
        "\n",
        "  # Filter blobs by area:\n",
        "  if rectArea > minArea:\n",
        "\n",
        "    # Draw bounding box:\n",
        "    color = (255, 0, 255)\n",
        "    cv2.rectangle(img_copy, (int(rectX), int(rectY)),\n",
        "                  (int(rectX + rectWidth), int(rectY + rectHeight)), color, 1)\n",
        "\n",
        "    # Crop bounding box:\n",
        "    currentCrop = img[rectY-3:rectY+rectHeight+3,rectX-3:rectX+rectWidth+3]\n",
        "    current_crop.append(currentCrop)\n",
        "    # Resize image to (45,45)\n",
        "    test_image = cv2.resize(currentCrop, (45,45), interpolation = cv2.INTER_AREA)\n",
        "    #change to PIL format\n",
        "    # test_image  = cropped_image\n",
        "    test_image = image.img_to_array(test_image) \n",
        "    test_image = np.expand_dims(test_image , axis = 0) #extra dimension to batch\n",
        "    result = global_model.predict(test_image)\n",
        "    result = lb.inverse_transform(result)\n",
        "    result1 = \"\".join(str(e) for e in result)\n",
        "    predictions.append(result1)\n",
        "    print(predictions)\n",
        "    cv2.putText(img_copy ,result1, (int(rectX), int(rectY)-5 ),cv2.FONT_HERSHEY_SIMPLEX,1,(0,0,255),1)\n",
        "    cv2_imshow(currentCrop)\n",
        "    # img_copy = cv2.resize(img_copy, (100,100), interpolation = cv2.INTER_AREA)\n",
        "    # cv2_imshow(img_copy)\n",
        "    plt.imshow(img_copy, cmap = 'gray')\n",
        "    plt.show()\n",
        "\n",
        "    cv2.waitKey(0)\n",
        "pred = \"\".join(str(e) for e in predictions)\n",
        "print(eval(pred))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "i5KQJ-PNf31w"
      },
      "source": [
        "## Model Evaluation Metrics\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cGHDx_4p9WBy",
        "outputId": "6dc578fe-f0b8-43bd-f45d-1abe58fd69b4"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "[[1.62797875e-09 1.34542990e-08 6.08463836e-07 ... 1.50830513e-08\n",
            "  8.72028036e-13 9.47508131e-12]\n",
            " [2.33303144e-04 9.08543766e-01 4.95233544e-05 ... 6.23217147e-06\n",
            "  4.78536967e-05 2.55212854e-07]\n",
            " [3.73847166e-13 6.90111923e-09 4.96726306e-11 ... 1.04268620e-04\n",
            "  7.82541133e-12 5.85806794e-12]\n",
            " ...\n",
            " [2.08677022e-08 4.83069243e-07 1.06348692e-04 ... 5.15243388e-04\n",
            "  1.07109925e-10 5.27016208e-11]\n",
            " [9.50727148e-23 1.29899086e-12 2.13343943e-11 ... 2.65481456e-11\n",
            "  9.87003054e-21 2.04515091e-20]\n",
            " [2.13788386e-18 6.09139974e-08 2.90005487e-13 ... 2.69836491e-08\n",
            "  4.84405708e-14 7.58452952e-15]]\n",
            "['-' '(' 'X' ... '2' 'sin' 'b']\n"
          ]
        }
      ],
      "source": [
        "y_pred  = global_model.predict(X_test)\n",
        "# y_pred = np.argmax(y_pred,axis=1)\n",
        "print(y_pred)\n",
        "y_pred = lb.inverse_transform(y_pred)\n",
        "print(y_pred)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9zgldhxd9xY9",
        "outputId": "601e8edf-7a85-4b52-e589-2e0ef5ed275d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "[[0 0 0 ... 0 0 0]\n",
            " [0 1 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]\n",
            " ...\n",
            " [0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]\n",
            " [0 0 0 ... 0 0 0]]\n",
            "['-' '(' 'times' ... '2' 'sin' 'b']\n"
          ]
        }
      ],
      "source": [
        "y_test1 = y_test.copy()\n",
        "print(y_test1)\n",
        "# y_test1 = lb.inverse_transform(y_test1)\n",
        "y_test1 = lb.inverse_transform(y_test1)\n",
        "print(y_test1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1iFRxqsyf7rR",
        "outputId": "1d6aa44d-3ed7-4c81-a23e-c2d3bd243a2d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(81, 81)\n"
          ]
        }
      ],
      "source": [
        "# Confusion Matrix\n",
        "from sklearn.metrics import confusion_matrix\n",
        "cm = confusion_matrix(y_test1,y_pred)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "H9ryRTSs9hmE"
      },
      "outputs": [],
      "source": [
        "import warnings\n",
        "warnings.filterwarnings('always')\n",
        "from sklearn.metrics import precision_recall_fscore_support"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "duuU9k9_Xhw_",
        "outputId": "c7a42ec7-5d6c-4cb0-9209-7561ae8be61b"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Precision: 0.901\n",
            " Recall: 0.909\n",
            " fscore: 0.902\n",
            " Support: None\n",
            "\n"
          ]
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/sklearn/metrics/_classification.py:1318: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
            "  _warn_prf(average, modifier, msg_start, len(result))\n"
          ]
        }
      ],
      "source": [
        "precision , recall , fscore , support = precision_recall_fscore_support(y_test1,y_pred , average='weighted')\n",
        "print(\"Precision: {:.3f}\\n Recall: {:.3f}\\n fscore: {:.3f}\\n Support: {}\\n\".format(precision , recall , fscore , support))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "yab4zs_IfLEA"
      },
      "outputs": [],
      "source": [
        "labels = ['!','(',')','+',',','-','0','1','2','3','4','5','6','7','8','9','=','A','C',\n",
        " 'Delta','G','H','M','N','R','S','T','X','[',']','alpha','ascii_124','b',\n",
        " 'beta','cos','d','div','e','exists','f','forall','forward_slash','gamma','geq','gt',\n",
        " 'i','in','infty','int','j','k','l','lambda','ldots','leq','lim','log','lt','mu',\n",
        " 'neq','o','p','phi','pi','pm',\n",
        "#  'prime',\n",
        " 'q','rightarrow','sigma','sin','sqrt','sum','tan','theta','times','u','v','w','y','z','{','}']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "CKgLa3GGZsRl"
      },
      "outputs": [],
      "source": [
        "# import seaborn as sns\n",
        "# import pandas as pd\n",
        "# import matplotlib.pyplot as plt\n",
        "\n",
        "# df_cm = pd.DataFrame(cm, index = [i for i in labels],\n",
        "#                   columns = [i for i in labels])\n",
        "# plt.figure(figsize = (30,30))\n",
        "# sns.heatmap(df_cm, annot=True , cmap = \"Blues\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "nFwzTBMZihMg",
        "outputId": "8e48178b-fbc5-4afc-befb-0d7895183c49"
      },
      "outputs": [
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x1440 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.figure(figsize = (20,20))\n",
        "plt.imshow(cm, cmap=plt.cm.Blues_r)\n",
        "plt.xlabel(\"Predicted labels\")\n",
        "plt.ylabel(\"True labels\")\n",
        "plt.xticks([], [])\n",
        "plt.yticks([], [])\n",
        "plt.title('Confusion matrix ')\n",
        "plt.colorbar()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "aUkQw_xGCPHi"
      },
      "outputs": [],
      "source": [
        "# import numpy as np\n",
        "# import matplotlib.pyplot as plt\n",
        "# from matplotlib.animation import FuncAnimation\n",
        "# from IPython.display import HTML\n",
        "# x_data = []\n",
        "# y_data = []\n",
        "\n",
        "# fig, ax = plt.subplots()\n",
        "# ax.set_xlim(0,20)\n",
        "# ax.set_ylim(0, 0.95)\n",
        "# line, = ax.plot(0, 0)\n",
        "# arr = list(range(0, len(global_acc_list)))\n",
        "# def animation_frame(i):\n",
        "#   x_data = arr[:i]\n",
        "#   y_data = global_acc_list[:i]\n",
        "\n",
        "#   line.set_xdata(x_data)\n",
        "#   line.set_ydata(y_data)\n",
        "\t\n",
        "#   return line, \n",
        "# animation = FuncAnimation(fig, func=animation_frame, frames = 100, interval=40 , blit= True)\n",
        "# HTML(animation.to_html5_video())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "id": "yZN-NQ9xYaj1",
        "outputId": "9c81ff7d-ee55-4c11-8ba2-a8f17efe01ca"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "IID | total comm rounds 20\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x288 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# IID \n",
        "import matplotlib.pyplot as plt\n",
        "plt.figure(figsize=(15,4))\n",
        "plt.subplot(121)\n",
        "plt.title(\"IID | total comm rounds: {}\".format(len(global_acc_list)))\n",
        "plt.plot(list(range(0,len(global_loss_list))), global_loss_list ,  label = \"global_loss\") \n",
        "plt.legend()\n",
        "plt.subplot(122)\n",
        "plt.title(\"IID | total comm rounds: {}\".format(len(global_acc_list)))\n",
        "plt.plot(list(range(0,len(global_acc_list))), global_acc_list , label = 'global_accuracy' , color = 'orange')\n",
        "plt.legend()\n",
        "print('IID | total comm rounds', len(global_acc_list))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "UF10q5RjYaj2"
      },
      "outputs": [],
      "source": [
        "iid_df = pd.DataFrame(list(zip(global_acc_list, global_loss_list)), columns =['global_acc_list', 'global_loss_list'])\n",
        "iid_df.to_csv('IID.csv',index=False)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EDxBPCz0Yaj3"
      },
      "source": [
        "### Non-IID"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "2ULRv0iKlWRd"
      },
      "outputs": [],
      "source": [
        "def batch_data(data_shard, bs=32):\n",
        "    '''Takes in a clients data shard and create a tfds object off it\n",
        "    args:\n",
        "        shard: a data, label constituting a client's data shard\n",
        "        bs:batch size\n",
        "    return:\n",
        "        tfds object'''\n",
        "    #seperate shard into data and labels lists\n",
        "    data, label = zip(*data_shard)\n",
        "    # dataset = tf.data.Dataset.from_tensor_slices(list(data), list(label))\n",
        "    dataset = tf.data.Dataset.from_tensor_slices((list(data), list(label)))\n",
        "    return dataset.shuffle(len(label)).batch(bs)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "JfvnvItPYaj4"
      },
      "outputs": [],
      "source": [
        "def create_clients(image_list, label_list, num_clients=100, initial='clients'):\n",
        "  ''' return: a dictionary with keys clients' names and value as \n",
        "              data shards - tuple of images and label lists.\n",
        "      args: \n",
        "          image_list: a list of numpy arrays of training images\n",
        "          label_list:a list of binarized labels for each image\n",
        "          num_client: number of fedrated members (clients)\n",
        "          initials: the clients'name prefix, e.g, clients_1 \n",
        "          \n",
        "  '''\n",
        "\n",
        "  #create a list of client names\n",
        "  client_names = ['{}_{}'.format(initial, i+1) for i in range(num_clients)]\n",
        "\n",
        "  #randomize the data\n",
        "  # data = list(zip(image_list, label_list))\n",
        "  # random.shuffle(data)  # <- IID\n",
        "  \n",
        "  # sort data for non-iid\n",
        "  max_y = np.argmax(label_list, axis=1)\n",
        "  sorted_zip = sorted(zip(max_y, label_list, image_list), key=lambda x: x[0])\n",
        "  data = [(x,y) for _,y,x in sorted_zip]\n",
        "\n",
        "  #shard data and place at each client\n",
        "  size = len(data)//num_clients\n",
        "  shards = [data[i:i + size] for i in range(0, size*num_clients, size)]\n",
        "\n",
        "  #number of clients must equal number of shards\n",
        "  # assert(len(shards) == len(client_names))\n",
        "\n",
        "  return {client_names[i] : shards[i] for i in range(len(client_names))} "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bhSwtmMqYaj6",
        "outputId": "73369599-0ccb-4e7d-badf-c7fbdd5742f7"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "(67644, 7517, 67644, 7517)"
            ]
          },
          "execution_count": 63,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "len(X_train), len(X_test), len(y_train), len(y_test)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "2WRvvt1uYaj6"
      },
      "outputs": [],
      "source": [
        "#create clients\n",
        "clients = create_clients(X_train, y_train, num_clients=10, initial='client')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "lS4Hr2cSYaj7"
      },
      "outputs": [],
      "source": [
        "#process and batch the training data for each client\n",
        "clients_batched = dict()\n",
        "for (client_name, data) in clients.items():\n",
        "    clients_batched[client_name] = batch_data(data)\n",
        "    \n",
        "#process and batch the test set  \n",
        "test_batched = tf.data.Dataset.from_tensor_slices((X_test, y_test)).batch(len(y_test))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "HuL63pOIYaj7"
      },
      "outputs": [],
      "source": [
        "comms_round = 20\n",
        "loss='categorical_crossentropy'\n",
        "metrics = ['accuracy']\n",
        "optimizer  = 'adam'\n",
        "  "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dyz5qkwUYaj7"
      },
      "outputs": [],
      "source": [
        "#initialize global model\n",
        "\n",
        "build_shape = [45,45,3] # Shape of Image\n",
        "classes = 82\n",
        "global_model = create_keras_model(build_shape , classes)\n",
        "global_acc_list = []\n",
        "global_loss_list = []"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "T5MNwlaCYaj8",
        "outputId": "83dad4fa-2b8a-452c-aa87-5fcbfc8ed79d"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3563 - accuracy: 0.8940\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2770 - accuracy: 0.9109\n",
            "212/212 [==============================] - 3s 11ms/step - loss: 0.2929 - accuracy: 0.9141\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3533 - accuracy: 0.8739\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 1.1590 - accuracy: 0.6348\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.8608 - accuracy: 0.7442\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.6784 - accuracy: 0.8004\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 1.3600 - accuracy: 0.6200\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 1.7945 - accuracy: 0.5171\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 1.1777 - accuracy: 0.6600\n",
            "comm_round: 0 | global_acc: 5.654% | global_loss: 4.40455436706543\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2286 - accuracy: 0.9450\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1556 - accuracy: 0.9630\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1475 - accuracy: 0.9691\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1993 - accuracy: 0.9446\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.8036 - accuracy: 0.7640\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.6738 - accuracy: 0.8088\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.5049 - accuracy: 0.8514\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 1.0753 - accuracy: 0.6946\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 1.3261 - accuracy: 0.6438\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.7764 - accuracy: 0.7812\n",
            "comm_round: 1 | global_acc: 24.385% | global_loss: 4.3903117179870605\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1491 - accuracy: 0.9626\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1010 - accuracy: 0.9759\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0961 - accuracy: 0.9809\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1260 - accuracy: 0.9675\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.5883 - accuracy: 0.8241\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.5014 - accuracy: 0.8547\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4256 - accuracy: 0.8800\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.8791 - accuracy: 0.7558\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 1.1081 - accuracy: 0.7012\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.6526 - accuracy: 0.8124\n",
            "comm_round: 2 | global_acc: 28.150% | global_loss: 4.3578901290893555\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1273 - accuracy: 0.9698\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0741 - accuracy: 0.9820\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0779 - accuracy: 0.9831\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0993 - accuracy: 0.9755\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.5189 - accuracy: 0.8517\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.4236 - accuracy: 0.8777\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3506 - accuracy: 0.8956\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.7564 - accuracy: 0.7853\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.9458 - accuracy: 0.7416\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.5338 - accuracy: 0.8474\n",
            "comm_round: 3 | global_acc: 33.112% | global_loss: 4.304141521453857\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1107 - accuracy: 0.9740\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0617 - accuracy: 0.9848\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0785 - accuracy: 0.9824\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0872 - accuracy: 0.9796\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4355 - accuracy: 0.8732\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3846 - accuracy: 0.8860\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3109 - accuracy: 0.9095\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.6512 - accuracy: 0.8195\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.8208 - accuracy: 0.7737\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4745 - accuracy: 0.8640\n",
            "comm_round: 4 | global_acc: 34.947% | global_loss: 4.266790390014648\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0983 - accuracy: 0.9777\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0547 - accuracy: 0.9861\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0565 - accuracy: 0.9877\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0682 - accuracy: 0.9820\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3814 - accuracy: 0.8876\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3480 - accuracy: 0.8993\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.2658 - accuracy: 0.9242\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.6016 - accuracy: 0.8289\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.7629 - accuracy: 0.7899\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4066 - accuracy: 0.8878\n",
            "comm_round: 5 | global_acc: 41.027% | global_loss: 4.196011066436768\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0808 - accuracy: 0.9802\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0470 - accuracy: 0.9880\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0489 - accuracy: 0.9899\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0570 - accuracy: 0.9851\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.3496 - accuracy: 0.8956\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3132 - accuracy: 0.9063\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2376 - accuracy: 0.9326\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.5424 - accuracy: 0.8446\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.6724 - accuracy: 0.8121\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3798 - accuracy: 0.8903\n",
            "comm_round: 6 | global_acc: 44.845% | global_loss: 4.159544467926025\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0763 - accuracy: 0.9830\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0458 - accuracy: 0.9885\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0446 - accuracy: 0.9902\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0507 - accuracy: 0.9871\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3143 - accuracy: 0.9061\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2858 - accuracy: 0.9111\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2117 - accuracy: 0.9385\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.5093 - accuracy: 0.8547\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.6293 - accuracy: 0.8250\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3377 - accuracy: 0.9024\n",
            "comm_round: 7 | global_acc: 47.186% | global_loss: 4.1432976722717285\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0680 - accuracy: 0.9834\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0407 - accuracy: 0.9910\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0468 - accuracy: 0.9901\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0490 - accuracy: 0.9870\n",
            "212/212 [==============================] - 4s 8ms/step - loss: 0.2844 - accuracy: 0.9148\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2490 - accuracy: 0.9212\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1990 - accuracy: 0.9426\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4753 - accuracy: 0.8637\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.5788 - accuracy: 0.8371\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2958 - accuracy: 0.9137\n",
            "comm_round: 8 | global_acc: 49.647% | global_loss: 4.101800441741943\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0687 - accuracy: 0.9843\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0334 - accuracy: 0.9917\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0352 - accuracy: 0.9923\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0488 - accuracy: 0.9868\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2719 - accuracy: 0.9185\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2475 - accuracy: 0.9208\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1850 - accuracy: 0.9443\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.4570 - accuracy: 0.8668\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.5307 - accuracy: 0.8510\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2672 - accuracy: 0.9245\n",
            "comm_round: 9 | global_acc: 54.038% | global_loss: 4.072140216827393\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0585 - accuracy: 0.9863\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0378 - accuracy: 0.9911\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0322 - accuracy: 0.9936\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0420 - accuracy: 0.9886\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.2594 - accuracy: 0.9221\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2397 - accuracy: 0.9209\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1820 - accuracy: 0.9468\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.4128 - accuracy: 0.8814\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.5419 - accuracy: 0.8514\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.2580 - accuracy: 0.9249\n",
            "comm_round: 10 | global_acc: 51.856% | global_loss: 4.063697814941406\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0569 - accuracy: 0.9863\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0387 - accuracy: 0.9904\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0365 - accuracy: 0.9923\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0404 - accuracy: 0.9886\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2298 - accuracy: 0.9295\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2149 - accuracy: 0.9339\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1657 - accuracy: 0.9525\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4050 - accuracy: 0.8785\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4935 - accuracy: 0.8630\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2472 - accuracy: 0.9268\n",
            "comm_round: 11 | global_acc: 55.594% | global_loss: 4.052562713623047\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0575 - accuracy: 0.9867\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0367 - accuracy: 0.9919\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0377 - accuracy: 0.9929\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0378 - accuracy: 0.9897\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2121 - accuracy: 0.9363\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2194 - accuracy: 0.9302\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1627 - accuracy: 0.9540\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3683 - accuracy: 0.8922\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.4815 - accuracy: 0.8658\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2270 - accuracy: 0.9351\n",
            "comm_round: 12 | global_acc: 58.268% | global_loss: 4.021512985229492\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0519 - accuracy: 0.9867\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0342 - accuracy: 0.9931\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0332 - accuracy: 0.9933\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0390 - accuracy: 0.9902\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1986 - accuracy: 0.9385\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1939 - accuracy: 0.9388\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1637 - accuracy: 0.9522\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3546 - accuracy: 0.8964\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.4499 - accuracy: 0.8729\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.2219 - accuracy: 0.9332\n",
            "comm_round: 13 | global_acc: 57.496% | global_loss: 4.017014026641846\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0507 - accuracy: 0.9886\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0345 - accuracy: 0.9923\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0313 - accuracy: 0.9931\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0378 - accuracy: 0.9901\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2074 - accuracy: 0.9354\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1907 - accuracy: 0.9413\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1605 - accuracy: 0.9512\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3347 - accuracy: 0.8999\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4021 - accuracy: 0.8884\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2161 - accuracy: 0.9358\n",
            "comm_round: 14 | global_acc: 60.184% | global_loss: 4.003483295440674\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0457 - accuracy: 0.9891\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0288 - accuracy: 0.9935\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0295 - accuracy: 0.9936\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0341 - accuracy: 0.9898\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1944 - accuracy: 0.9450\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1733 - accuracy: 0.9483\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1443 - accuracy: 0.9555\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.3270 - accuracy: 0.9014\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.4127 - accuracy: 0.8788\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2079 - accuracy: 0.9366\n",
            "comm_round: 15 | global_acc: 63.603% | global_loss: 3.967172622680664\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0359 - accuracy: 0.9910\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0309 - accuracy: 0.9922\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0299 - accuracy: 0.9923\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0326 - accuracy: 0.9908\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1829 - accuracy: 0.9440\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1773 - accuracy: 0.9462\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1398 - accuracy: 0.9595\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.3307 - accuracy: 0.9014\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.3717 - accuracy: 0.8944\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1912 - accuracy: 0.9423\n",
            "comm_round: 16 | global_acc: 64.840% | global_loss: 3.942953109741211\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0361 - accuracy: 0.9905\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0321 - accuracy: 0.9931\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0272 - accuracy: 0.9944\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.0321 - accuracy: 0.9905\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1703 - accuracy: 0.9506\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1708 - accuracy: 0.9450\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1366 - accuracy: 0.9586\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3001 - accuracy: 0.9103\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3665 - accuracy: 0.8965\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1910 - accuracy: 0.9440\n",
            "comm_round: 17 | global_acc: 66.343% | global_loss: 3.94997239112854\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0438 - accuracy: 0.9901\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0324 - accuracy: 0.9939\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.0258 - accuracy: 0.9948\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0287 - accuracy: 0.9933\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1623 - accuracy: 0.9494\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1652 - accuracy: 0.9491\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.1365 - accuracy: 0.9619\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.2883 - accuracy: 0.9147\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.3517 - accuracy: 0.8955\n",
            "212/212 [==============================] - 3s 9ms/step - loss: 0.1681 - accuracy: 0.9487\n",
            "comm_round: 18 | global_acc: 68.964% | global_loss: 3.916574001312256\n",
            "['client_1', 'client_2', 'client_3', 'client_4', 'client_5', 'client_6', 'client_7', 'client_8', 'client_9', 'client_10']\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0418 - accuracy: 0.9894\n",
            "212/212 [==============================] - 4s 8ms/step - loss: 0.0298 - accuracy: 0.9945\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0309 - accuracy: 0.9936\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.0344 - accuracy: 0.9917\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1700 - accuracy: 0.9514\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1535 - accuracy: 0.9493\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1200 - accuracy: 0.9627\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.2859 - accuracy: 0.9182\n",
            "212/212 [==============================] - 2s 8ms/step - loss: 0.3440 - accuracy: 0.9001\n",
            "212/212 [==============================] - 3s 8ms/step - loss: 0.1575 - accuracy: 0.9514\n",
            "comm_round: 19 | global_acc: 69.017% | global_loss: 3.910620927810669\n"
          ]
        }
      ],
      "source": [
        "#commence global training loop\n",
        "for comm_round in range(comms_round):\n",
        "            \n",
        "  # get the global model's weights - will serve as the initial weights for all local models\n",
        "  global_weights = global_model.get_weights()\n",
        "\n",
        "  #initial list to collect local model weights after scalling\n",
        "  scaled_local_weight_list = list()\n",
        "\n",
        "  #randomize client data - using keys\n",
        "  # all_client_names = list(clients_batched.keys())\n",
        "          \n",
        "  client_names = list(clients_batched.keys())\n",
        "  print(client_names)\n",
        "  # random.shuffle(client_names)\n",
        "\n",
        "  #loop through each client and create new local model\n",
        "  for client in client_names:\n",
        "    local_model = create_keras_model(build_shape, classes)\n",
        "    local_model.compile(loss=loss, \n",
        "                optimizer=optimizer, \n",
        "                metrics=metrics)\n",
        "\n",
        "    #set local model weight to the weight of the global model\n",
        "    local_model.set_weights(global_weights)\n",
        "\n",
        "    #fit local model with client's data\n",
        "    local_model.fit(clients_batched[client], epochs=1, workers = 8 )\n",
        "      \n",
        "    #scale the model weights and add to list\n",
        "    scaling_factor = 0.1 # weight_scalling_factor(clients_batched, client)\n",
        "    scaled_weights = scale_model_weights(local_model.get_weights(), scaling_factor)\n",
        "    scaled_local_weight_list.append(scaled_weights)\n",
        "    # print(scaled_local_weight_list)\n",
        "      \n",
        "    #clear session to free memory after each communication round\n",
        "    K.clear_session()\n",
        "      \n",
        "  #to get the average over all the local model, we simply take the sum of the scaled weights\n",
        "  average_weights = sum_scaled_weights(scaled_local_weight_list)\n",
        "\n",
        "  #update global model \n",
        "  global_model.set_weights(average_weights)\n",
        "\n",
        "  #test global model and print out metrics after each communications round\n",
        "  for(X_test, Y_test) in test_batched:\n",
        "      global_acc, global_loss = test_model(X_test, Y_test, global_model, comm_round)\n",
        "      global_acc_list.append(global_acc)\n",
        "      global_loss_list.append(global_loss)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "id": "E0kP3CxAYaj-",
        "outputId": "705b53f0-eb72-49a6-e7ee-53f76c52b581"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "NON-IID | total comm rounds 20\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x288 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# NON-IID \n",
        "import matplotlib.pyplot as plt\n",
        "plt.figure(figsize=(15,4))\n",
        "plt.subplot(121)\n",
        "plt.title(\"NON-IID | total comm rounds: {}\".format(len(global_acc_list)))\n",
        "plt.plot(list(range(0,len(global_loss_list))), global_loss_list ,  label = \"global_loss\") \n",
        "plt.legend()\n",
        "plt.subplot(122)\n",
        "plt.title(\"NON-IID | total comm rounds: {}\".format(len(global_acc_list)))\n",
        "plt.plot(list(range(0,len(global_acc_list))), global_acc_list , label = 'global_accuracy' , color = 'orange')\n",
        "plt.legend()\n",
        "print('NON-IID | total comm rounds', len(global_acc_list))"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Traing By Solely on CNN for 20 Epochs"
      ],
      "metadata": {
        "id": "kYK4aoe3BgSl"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YPT8KoHSvq5p"
      },
      "outputs": [],
      "source": [
        "cnn_dataset = tf.data.Dataset.from_tensor_slices((X_train, y_train)).batch(32)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "hSQPXYIOwLX6"
      },
      "outputs": [],
      "source": [
        "build_shape = [45,45,3] # Train image Dimensions\n",
        "classes = 82\n",
        "loss='categorical_crossentropy'\n",
        "metrics = ['accuracy']\n",
        "optimizer  = 'adam'   "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "PQGRIxnEv99K"
      },
      "outputs": [],
      "source": [
        "cnn_model = create_keras_model( build_shape , classes)\n",
        "cnn_model.compile(\n",
        "  loss=loss, \n",
        "  optimizer=optimizer, \n",
        "  metrics=metrics\n",
        ")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "background_save": true,
          "base_uri": "https://localhost:8080/"
        },
        "id": "jD5VL82OxIdq",
        "outputId": "13d65aa0-3452-4a2d-faa4-d16c18f0ee30"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "2114/2114 [==============================] - 202s 94ms/step - loss: 1.3969 - accuracy: 0.6450\n",
            "Epoch 2/20\n",
            "2114/2114 [==============================] - 143s 66ms/step - loss: 0.7519 - accuracy: 0.7924\n",
            "Epoch 3/20\n",
            "2114/2114 [==============================] - 143s 66ms/step - loss: 0.6084 - accuracy: 0.8262\n",
            "Epoch 4/20\n",
            "2114/2114 [==============================] - 144s 67ms/step - loss: 0.5286 - accuracy: 0.8472\n",
            "Epoch 5/20\n",
            "2114/2114 [==============================] - 144s 67ms/step - loss: 0.4783 - accuracy: 0.8609\n",
            "Epoch 6/20\n",
            "2114/2114 [==============================] - 144s 67ms/step - loss: 0.4344 - accuracy: 0.8708\n",
            "Epoch 7/20\n",
            "2114/2114 [==============================] - 145s 67ms/step - loss: 0.3990 - accuracy: 0.8808\n",
            "Epoch 8/20\n",
            "2114/2114 [==============================] - 145s 67ms/step - loss: 0.3709 - accuracy: 0.8884\n",
            "Epoch 9/20\n",
            "2114/2114 [==============================] - 144s 67ms/step - loss: 0.3490 - accuracy: 0.8938\n",
            "Epoch 10/20\n",
            "2114/2114 [==============================] - 146s 68ms/step - loss: 0.3269 - accuracy: 0.9001\n",
            "Epoch 11/20\n",
            "2114/2114 [==============================] - 146s 67ms/step - loss: 0.3151 - accuracy: 0.9032\n",
            "Epoch 12/20\n",
            "2114/2114 [==============================] - 146s 68ms/step - loss: 0.2987 - accuracy: 0.9069\n",
            "Epoch 13/20\n",
            "2114/2114 [==============================] - 146s 68ms/step - loss: 0.2819 - accuracy: 0.9118\n",
            "Epoch 14/20\n",
            "2114/2114 [==============================] - 148s 68ms/step - loss: 0.2723 - accuracy: 0.9157\n",
            "Epoch 15/20\n",
            "2114/2114 [==============================] - 148s 68ms/step - loss: 0.2689 - accuracy: 0.9159\n",
            "Epoch 16/20\n",
            "2114/2114 [==============================] - 147s 68ms/step - loss: 0.2548 - accuracy: 0.9198\n",
            "Epoch 17/20\n",
            "2114/2114 [==============================] - 147s 68ms/step - loss: 0.2476 - accuracy: 0.9212\n",
            "Epoch 18/20\n",
            "2114/2114 [==============================] - 147s 68ms/step - loss: 0.2439 - accuracy: 0.9240\n",
            "Epoch 19/20\n",
            "2114/2114 [==============================] - 148s 68ms/step - loss: 0.2345 - accuracy: 0.9257\n",
            "Epoch 20/20\n",
            "2114/2114 [==============================] - 147s 68ms/step - loss: 0.2269 - accuracy: 0.9282\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f6659e66210>"
            ]
          },
          "execution_count": null,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "cnn_model.fit(cnn_dataset , epochs = 20)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "ubbOlqiIxc4j"
      },
      "outputs": [],
      "source": [
        "# for(X_test, Y_test) in test_batched:\n",
        "#         acc, loss = test_model(X_test, Y_test, cnn_model, 1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "bu--HmHm66Mr"
      },
      "outputs": [],
      "source": [
        "acc_list = cnn_model.history.history['accuracy']\n",
        "loss_list = cnn_model.history.history['loss']\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "id": "EqkLw3US-Ng4",
        "outputId": "a15c294b-fa8c-4885-df81-37727ca472b4"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "CNN Training for 20 epochs\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x288 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Single CNN \n",
        "import matplotlib.pyplot as plt\n",
        "plt.figure(figsize=(15,4))\n",
        "plt.subplot(121)\n",
        "plt.title(\"CNN Training Losses for 20 epochs\")\n",
        "plt.plot(list(range(0,len(loss_list))), loss_list ,  label = \"CNN_loss\") \n",
        "plt.legend()\n",
        "plt.subplot(122)\n",
        "plt.title(\"CNN Training Accuracy for 20 epochs\")\n",
        "plt.plot(list(range(0,len(acc_list))), acc_list , label = 'CNN_accuracy' , color = 'orange')\n",
        "plt.legend()\n",
        "print('CNN Training for 20 epochs')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "65gGYkoW7ECj"
      },
      "outputs": [],
      "source": [
        "y_pred  = cnn_model.predict(X_test)\n",
        "# y_pred = np.argmax(y_pred,axis=1)\n",
        "print(y_pred)\n",
        "y_pred = lb.inverse_transform(y_pred)\n",
        "print(y_pred)"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "name": "federated-learning Maths Handwritten.ipynb",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.9.7"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}